From: Prognostic models for complete recovery in ischemic stroke: a systematic review and meta-analysis
Assessment items | All models (n = 23) |
---|---|
Study design | |
Cohort study | 23 (100%) |
Variables | |
Description of measurement of predictors | |
 Yes | 13 (56.5%) |
 No | 10 (43.5%) |
Loss to follow-up | Â |
  < 10% | 10 (43.5%) |
  ≥ 10% | 13 (56.5%) |
Analysis | |
More than 10 events per variable | |
 Yes | 22 (95.7%) |
 No | 1 (4.3%) |
Method for selection of predictors during multivariable modeling | |
 Forward Selection | 2 (8.7%) |
 Backward Elimination | 3 (13.0%) |
 Stepwise selection | 0 |
 Full model approach | 16 (69.6%) |
 Unknown | 2 (8.7%) |
Handling of missing data | Â |
 Estimated statistically | 0 |
 Excluded | 23 (100%) |
Model performance | |
Internal validity | |
Performance reported AUC (Discrimination) | |
 Yes | 12 (52.2%) |
 95% CI presented | 0 |
 No | 11 (47.8%) |
Calibration | |
 Yes | 1 (4.3) |
 No | 22 (95.7%) |
External validity | |
Performance reported AUC (Discrimination) | |
 Yes | 8 (34.9%) |
 95% CI presented | 2 out of 8 (25.0%) |
 No | 15 (65.1%) |
Calibration | |
 Yes | 6 (26.1%) |
 No | 17 (73.9%) |